Dynamic geothermal resource assessment: Integrating reservoir simulation and Gaussian Kernel Density Estimation under geological uncertainties

Abstract

This paper presents a dynamic geothermal resource assessment method that integrates reservoir simulation and Gaussian kernel density estimation (KDE). This approach addresses geological uncertainties by employing reservoir simulation techniques to model the fluid and heat flow under the condition of permeability heterogeneity. Incorporating probabilistic resource assessment through Gaussian KDE, the study quantifies uncertainties, estimating the probability density function (PDF) of ensemble results under conditions like thermal breakthrough thresholds, fixed reservoir lifespans, and target energy production. The demonstrations of assessment start with a simple homogeneous model. The results show that larger doublet well distances result in extended lifespan, higher final production well temperatures, and increased energy production. Brugge reservoir emphasizes the impact of heterogeneity and uncertainty on production outcomes, especially at smaller doublet well distances. Assessment of fluvial Egg model reveals that drilling in fluvial channels causes rapid thermal breakthrough. This result indicates that, to optimize reservoir performance, it is recommended to refrain from drilling doublet wells within high-permeability fluvial channels. Furthermore, it is worthy of mention that the Gaussian kernel is not always favored for KDE, particularly in scenarios involving non-Gaussian distribution ensembles. The proposed method, which integrates reservoir simulation and Gaussian KDE, enhances understanding of geological uncertainties and the intricate nature of geothermal reservoirs, facilitating more reliable and accurate resource assessments.